Publication number | US7228000 B2 |

Publication type | Grant |

Application number | US 10/390,193 |

Publication date | Jun 5, 2007 |

Filing date | Mar 14, 2003 |

Priority date | Mar 15, 2002 |

Fee status | Paid |

Also published as | US7577308, US20030228061, US20070189622 |

Publication number | 10390193, 390193, US 7228000 B2, US 7228000B2, US-B2-7228000, US7228000 B2, US7228000B2 |

Inventors | Hiroyuki Sakuyama |

Original Assignee | Ricoh Co., Ltd. |

Export Citation | BiBTeX, EndNote, RefMan |

Patent Citations (27), Non-Patent Citations (7), Referenced by (8), Classifications (12), Legal Events (3) | |

External Links: USPTO, USPTO Assignment, Espacenet | |

US 7228000 B2

Abstract

An image data generator for generating data on a size reduced image of an original image. In one embodiment, the image data generator includes: an inverse wavelet transform unit to perform on inverse wavelet transform on wavelet coefficients; and a coefficient selecting unit to select, of wavelet coefficients in a sub-band, wavelet coefficients to be subjected to the inverse wavelet transform. In one embodiment, the inverse wavelet transform unit performs the inverse wavelet transform only on the wavelet coefficients selected in the coefficient selecting unit with respect to the wavelet coefficients in the sub-band.

Claims(14)

1. An apparatus for generating data for a reduced size image of an original image, the apparatus comprising:

a coefficient selecting unit to select wavelet coefficients to be subjected to an inverse wavelet transform in a same sub-band from wavelet coefficients obtained by performing a wavelet transform on the original image one or more times; and

an inverse wavelet transform unit to perform the inverse wavelet transform only on the wavelet coefficients selected in the coefficient selecting unit with respect to the sub-band,

wherein the coefficient selecting unit selects the wavelet coefficients required to obtain desired pixels of the original image.

2. The apparatus as claimed in claim 1 , wherein the coefficient selecting unit comprises a unit to select at least all of wavelet coefficients relating to pixels of the size-reduced image to be generated.

3. The apparatus as claimed in claim 1 , wherein the coefficient selecting unit comprises a unit to select only wavelet coefficients relating to pixels of the size-reduced image to be generated.

4. The apparatus as claimed in claim 1 , wherein the inverse wavelet transform comprises a plurality of inverse wavelet transforms; and

the coefficient selecting unit comprises a unit to select, of the wavelet coefficients in the sub-band in an interleaved state, all of wavelet coefficients positioned on straight lines extending in a direction in which the last one of the inverse wavelet transforms is performed, the straight lines each including a wavelet coefficient corresponding to a pixel position of the size-reduced image to be generated.

5. The apparatus as claimed in claim 1 , wherein:

the inverse wavelet transform is performed two-dimensionally in first and second directions perpendicular to each other in an order; and

the coefficient selecting unit comprises a unit that, in selecting wavelet coefficients to be subjected to the inverse wavelet transform performed in the first direction, selects, of the wavelet coefficients in the sub-band in an interleaved state, all of wavelet coefficients positioned on straight lines extending in the second direction, the straight lines including a wavelet coefficient corresponding to a pixel position of the size-reduced image to be generated or a wavelet coefficient required to obtain the pixel position by the inverse wavelet transform in the first direction.

6. The apparatus as claimed in claim 1 , wherein the coefficient selecting unit comprises a unit to select all of wavelet coefficients in sub-bands of a decomposition level higher than a decomposition level of an LL sub-band having a maximum resolution below a resolution of the size-reduced image to be generated, in order to obtain the LL sub-band by the inverse wavelet transform.

7. The apparatus as claimed in claim 1 , further comprising a high-pass filter and a low-pass filter having different tap lengths for the inverse wavelet transform, wherein the sum of the tap lengths of the high-pass and low-pass filters performing filtering centered on wavelet coefficients corresponding to pixel positions of the size-reduced image to be generated of the wavelet coefficients in the sub-band in an interleaved state is smaller than

*d·n*(*T*low+*T*high)/2

where d is a dimensional number of the inverse wavelet transform, n is the number of pixels of the size-reduced image, Tlow is the tap length of the low-pass filter, and Thigh is the tap length of the high-pass filter.

8. The apparatus as claimed in claim 1 , wherein the inverse wavelet transform is performed independently on each of a plurality of components with at least one of the components having a resolution set therefor, the resolution being set different from a resolution set for the rest of the components.

9. The apparatus as claimed in claim 8 , wherein:

the components comprises a luminance component and two color-difference components; and

the luminance component is subjected to the inverse wavelet transform to a resolution higher than the color-difference components.

10. An image data generator for generating data for a reduced size image of an original image, the image data generator comprising:

a coefficient selection unit to select wavelet coefficients to be subjected to an inverse wavelet transform in a same sub-band from wavelet coefficients obtained by performing a wavelet transform on the original image one or more times;

an inverse wavelet transform unit to perform inverse wavelet transform independently on each of the components with at least one of the components having a resolution set therefor, the resolution being set different from a resolution set for the rest of the components;

wherein the coefficient selecting unit selects the wavelet coefficients required to obtain desired pixels of the original image.

11. The image data generator as claimed in claim 10 , wherein:

the wavelet coefficients are of a plurality of components:

the components comprises a luminance component and two color-difference components; and

the luminance component is subjected to the inverse wavelet transform to a resolution higher than the color-difference components.

12. A computer-implemented method of generating a reduced size image of an original image, the method comprising:

(a) selecting wavelet coefficients to be subjected to an inverse wavelet transform in a same sub-band from wavelet coefficients obtained by performing a wavelet transform on the original image one or more times; and

(b) performing the inverse wavelet transform only on the wavelet coefficients selected to be subjected to the inverse wavelet transform with respect to the sub-band to generate data of the reduced size image,

wherein the coefficient selecting unit selects the wavelet coefficients required to obtain desired pixels of the original image.

13. An article of manufacture having one or more computer-readable recording media storing instructions which, when executed by a system, cause the system to generate data for a reduced size image of an original image by:
wherein the coefficient selecting unit selects the wavelet coefficients required to obtain desired pixels of the original image.

(a) selecting wavelet coefficients to be subjected to an inverse wavelet transform in a same sub-band from wavelet coefficients obtained by performing a wavelet transform on the original image one or more times; and

(b) performing the inverse wavelet transform only on the wavelet coefficients selected to be subjected to the inverse wavelet transform with respect to the sub-band to generate data of the reduced size image,

14. An apparatus of generating a reduced size image of an original image, the apparatus comprising:
wherein the coefficient selecting unit selects the wavelet coefficients required to obtain desired pixels of the original image.

means for selecting wavelet coefficients to be subjected to an inverse wavelet transform in a same sub-band from wavelet coefficients obtained by performing a wavelet transform on the original image one or more times; and

means for performing the inverse wavelet transform only on the wavelet coefficients selected to be subjected to an inverse wavelet transform with respect to the sub-band,

Description

The present application claims priority to the corresponding Japanese priority application 2002-072696 filed on Mar. 15, 2002, the entire contents of which are hereby incorporated by reference.

The present invention generally relates to image data generators, and more particularly to an image data generator for generating a reduced size image of an original. For purpose of this application, a reduced size image can be the one whose aspect ratio is different from that of the original image. The present invention is applicable to an application program for generating a reduced size image of an original, a device driver such as a printer driver, and other apparatuses for processing images.

Wavelet transforms have been employed more often lately as a substitute frequency conversion technique for DCTs (Discrete Cosine Transforms) employed in JPEG (Joint Photographic Coding Experts Group). A typical example of the use of wavelet transforms is JPEG 2000, which is an image compression and decompression method standardized internationally in 2001 as a successor to JPEG.

A wavelet transform is characterized in that an image is compressed at high compression rates with excellent image quality, and at the time of compression, is decomposed into a plurality of resolution components so that an image having a lower resolution than the original image is generated.

Conventionally, however, the following restriction exists in generating a low-resolution image from the original image. According to the conventional wavelet decoding entailing resolution conversion, resolution can be reduced only to 1/2^{n }(n=1, 2 . . . ) due to the nature of the wavelet decoding. This is because the conventional wavelet transform employs a two-division filter bank. Accordingly, the low-frequency components can be synthesized only in 1/2^{n }in a synthesis filter bank in the process of decoding, so that the reduction rate of the decoded image is limited to 1/2^{n}.

On the other hand, there is a demand for decoding with resolutions other than 1/2^{n }increases as the original image has a higher resolution. That is, if decoding can be performed with resolutions of any rational numbers including 1/2^{n}, the decoding is independent of the constraints on the terminal side, thereby considerably expanding its application range.

With respect to this resolution conversion, Japanese Laid-Open Patent Applications No. 2000-125293 and No. 2000-125294 (hereinafter referred to as first prior art and second prior art, respectively), showing the above-described problem in resolution conversion, disclose wavelet decoders and methods. Those wavelet decoders and methods are allowed to decode an image signal compressed and coded by wavelet transforms with resolutions of any rational numbers without being affected by the constraints on the terminal side. As a result, those wavelet decoders and methods are allowed to effectively store and display, for instance, a so-called thumbnail, which is frequently used in electronic still cameras and printers, and an image obtained by converting the resolution of an original image (a reduced or enlarged original image). Therefore, those wavelet decoders and methods can be applied to a wider variety of products.

Each of the wavelet decoders in the first prior art and the second prior art includes an entropy decoding unit that applies entropy decoding to a coded bit stream, an inverse quantization unit that inversely quantizes quantization coefficients and outputs transform coefficients, a transform coefficient inverse scanning unit that scans the transform coefficients according to a predetermined method and rearranges the transform coefficients, and a wavelet inverse transform unit that inversely transforms the rearranged transform coefficients and provides a decoded image. The wavelet inverse transform unit includes means for limiting the bandwidth of the transform coefficients in accordance with magnifications of resolution conversion, and has a configuration where up-samplers, down-samplers, and synthesis filers are adaptively arranged.

**119**, **121**, **124**, **126**, **129** and **131**, low-pass filters **120**, **125**, and **130** for synthesis, high-pass filters **122**, **127**, and **132** for synthesis, adders **123**, and a down-sampler **134**. Due to band restriction, the coefficients of resolution components (LH signal and H signal) **111** and **112** higher than one third of the resolution of the original image are not employed. On the other hand, the coefficient of low-resolution components (LLL signal and LLH signal) **109** and **110** are combined and output from the adder **123** further to be inversely converted, so that an image (signal) **114** whose resolution is a half of the resolution of the original image is obtained. The signal **114** is further inversely transformed so that an image (signal) **117** that has the same resolution as the original image is obtained. Finally, the signal **117** is thinned out or reduced to one third by the down-sampler **134** so that a desired image (signal) **118** is obtained. In

Thus, in both of the first prior art and the second prior art described above, it is not clear what “configuration where . . . are adaptively arranged” means. In the embodiments of those inventions, wavelet transform is performed after band restriction so that a decoded image is obtained. Thereafter, the decoded image is thinned out by the down-sampler provided at the final stage, so that a reduced image is obtained. Such a configuration, however, does not also perform inverse wavelet transform and decoding on pixels that are finally eliminated, and therefore, lacks efficiency

Japanese Laid-Open Patent Applications No. 2002-152517, No. 2002-152744, and No. 2002-344732 disclose an apparatus and methods for obtaining a decompressed image from a coded image. In these inventions, wavelet coefficients of up to a resolution level close to a desired size are extracted to be subjected to inverse wavelet transform so that a decoded image is generated. Thereafter, processing for varying magnification is performed so that an image of the desired size is obtained. In these inventions, it is required to vary magnification.

Further, the technique of imposing band restriction on a wavelet coefficient in the first prior art and the second prior art may be effective to some extent in increasing processing speed. This technique, however, is based on the premise that all of the codes of the wavelet coefficients that are not subjected to band restriction are decoded in the entropy decoding unit, and therefore, is not efficient.

An image data generation method and apparatus are disclosed. In one embodiment, an image data generator generates data of a reduced size image of an original image. The image data generator comprises an inverse wavelet transform unit to perform an inverse wavelet transform on wavelet coefficients and a coefficient selecting unit to select wavelet coefficients in a sub-band, wavelet coefficients to be subjected to the inverse wavelet transform. The inverse wavelet transform unit performs the inverse wavelet transform only on the wavelet coefficients selected in the coefficient selecting unit with respect to the wavelet coefficients in the sub-band.

Other objects, features and advantages of the present invention will become more apparent from the following detailed description when read in conjunction with the accompanying drawings, in which:

**2**LL sub-band, and

An image data generator and an image data generating method in which the above-described disadvantages are eliminated is disclosed.

Focusing on the fact that a wavelet coefficient inherently corresponds to its pixel position, one embodiment of the present invention provides an image data generator and an image data generating method that can generate decoded image data of a desired resolution, or desired reduced size image data, with a smaller amount of processing by performing inverse wavelet transform only on the coefficients that are related to pixels desired to be finally obtained, using the correspondence between the wavelet coefficients and their pixel positions. As mentioned above, a reduced size image can be the one whose aspect ratio is not equal to that of the original image.

Another embodiment of the present invention provides an image data generator and an image data generating method that can generate decoded image data of a desired resolution, or desired reduced size image data, balancing a decoding rate and decoding image quality, by timely performing partial decoding on coded wavelet coefficients.

Another embodiment of the present invention comprises image data generator for generating data on a reduced size image of an original image, where the image data generator includes an inverse wavelet transform unit to perform an inverse wavelet transform on wavelet coefficients and a coefficient selecting unit to select wavelet coefficients in a sub-band, wavelet coefficients to be subjected to the inverse wavelet transform, wherein the inverse wavelet transform unit performs the inverse wavelet transform only on the wavelet coefficients selected in the coefficient selecting unit with respect to the wavelet coefficients in the sub-band.

Another embodiment of the present invention comprises an image data generator for generating data on a reduced size image of an original image, where the image data generator includes an inverse wavelet transform unit to perform an inverse wavelet transform on wavelet coefficients of a plurality of components, the inverse wavelet transform unit to perform the inverse wavelet transform independently on each of the components with at least one of the components having a resolution set, which is different from a resolution set for the rest of the components.

Another embodiment of the present invention comprises an image data generator for generating data on a reduced size image of an original image, where the image data generator includes a coefficient decoding unit to decode codes into which wavelet coefficients are coded and a code omitting unit to omit codes not to be decoded of codes in a sub-band, where the coefficient decoding unit decodes codes other than those omitted by the code omitting unit.

According to the above-described image data generators of the present invention, the wavelet coefficients are subjected to inverse wavelet transform, and in the case of generating a reduced size image of an original image, the wavelet coefficients in the same sub-band are selectively subjected to an inverse wavelet transform. Therefore, the reduced size image can be generated with a reduced amount of calculation.

Further, the wavelet coefficients in the same sub-band are partially decoded in performing an inverse wavelet transform on the wavelet coefficients and generating a reduced size image of an original image. Therefore, the reduced size image can be generated with decoding speed and decoded image quality being balanced with each other.

In another embodiment of the present invention, a method of generating data on a reduced size image of an original image is disclosed, where the method includes: (a) selecting, of wavelet coefficients in a sub-band, wavelet coefficients to be subjected to an inverse wavelet transform; and (b) performing the inverse wavelet transform on wavelet coefficients, where performing the inverse wavelet transform includes performing the inverse wavelet transform only on the wavelet coefficients selected to be subjected to the inverse wavelet transform with respect to the wavelet coefficients in the sub-band.

Another embodiment of the present invention comprises a method of generating data on a reduced size image of an original image, where the method includes performing an inverse wavelet transform on wavelet coefficients of a plurality of components, wherein the inverse wavelet transform is performed independently on each of the components with at least one of the components having a resolution set which is different from a resolution set for the rest of the components. Another embodiment of the present invention includes a method of generating data on a reduced size image of an original image, where the method comprises: (a) decoding codes into which wavelet coefficients are coded; (b) omitting codes not to be decoded from codes in a sub-band; and (c) decoding codes other than those omitted as not to be decoded.

Another embodiment of the present invention comprises a computer-readable recording medium storing a program for causing a computer to execute a method of generating data on a reduced size image of an original image, where the method includes: (a) selecting, of wavelet coefficients in a sub-band, wavelet coefficients to be subjected to an inverse wavelet transform; and (b) performing the inverse wavelet transform on wavelet coefficients, where performing the inverse wavelet transform includes performing the inverse wavelet transform only on the wavelet coefficients selected to be subjected to the inverse wavelet transform with respect to the wavelet coefficients in the sub-band.

Another of the present invention comprises a computer-readable recording medium storing a program for causing a computer to execute a method of generating data on a reduced size image of an original image, where the method includes performing an inverse wavelet transform on wavelet coefficients of a plurality of components, wherein the inverse wavelet transform is performed independently on each of the components with at least one of the components having a resolution set therefor, the resolution being set different from a resolution set for the rest of the components. Another embodiment of the present invention comprises a computer-readable recording medium storing a program for causing a computer to execute a method of generating data on a reduced size image of an original image, where the method includes: (a) decoding codes into which wavelet coefficients are coded; (b) omitting codes not to be decoded of codes in a sub-band; and (c) decoding codes other than those omitted codes.

A description will now be given, with reference to the accompanying drawings, of embodiments of the present invention.

First, a description will be given, with respect to the present invention, of the reason a decoded image of a desired resolution can be generated with a reduced amount of processing by performing an inverse wavelet transform only on the coefficients that relate to pixels desired to be finally obtained, focusing on the correspondence between wavelet coefficients and their pixel positions.

In the following description, a vertical direction refers to upward and downward directions and a horizontal direction refers to rightward and leftward directions in the accompanying drawings.

As shown in

*C*(2*i+*1)=*P*(2*i+*1)−└(*P*(2*i*)+*P*(2*i+*2))/2┘ (1)

*C*(2*i*)=*P*(2*i*)+└(*C*(2*i−*1)+*C*(2*i+*1)+2)/4┘ (2)

where the symbol └x┘ represents a floor function of x that replaces a real number x with the largest of all the integers smaller than or equal to x. In the edge units of the image, a pixel in the center may lack an adjacent pixel. In this case, an appropriate pixel value is supplemented in accordance with a predetermined rule, which is irrelevant to the essence of the present invention so that a description thereof will be omitted.

If, for the purpose of simplification, the coefficients obtained by high-pass filtering are expressed as H, and the coefficients obtained by low-pass filtering are expressed as L, then a coefficient array **1** of **2** of H and L coefficients as shown in

Next, high-pass filtering is performed horizontally (in the positive direction of the X-axis) on each coefficient of an odd X-coordinate value x (x=2i+1) using its adjacent coefficients in the coefficient array **2** of

If, for the purpose of simplification, the coefficients obtained by performing low-pass filtering on the above-described coefficients L using its adjacent coefficients are expressed as LL, the coefficients obtained by performing high-pass filtering on the above-described coefficients L using its adjacent coefficients are expressed as HL, the coefficients obtained by performing low-pass filtering on the above-described coefficients H using its adjacent coefficients are expressed as LH, and the coefficients obtained by performing high-pass filtering on the above-described coefficients H using its adjacent coefficients are expressed as HH, then the coefficient array **2** of **3** of

Thus, one wavelet transform operation (one decomposition) is completed. At this point, only the LL coefficients are collected. That is, as a coefficient array **4** of **4** _{1}, HL sub-band **4** _{2}, LH sub-band **4** _{3}, and HH sub-band **4** _{4}), and only the LL sub-band **4** _{1 }is extracted. In so doing, a half-resolution “image” of the original image is obtained. Such grouping by the sub-band is referred to as deinterleaving, while the coefficients arranged in the state of **4** _{1 }as an original image and perform the same conversion as described above. In this case, when the coefficients are rearranged, a typical coefficient array **5** (of LL sub-band **5** _{1}, HL sub-band **5** _{2}, LH sub-band **5** _{3}, HH sub-band **5** _{4}, HL sub-band **4** _{2}, LH sub-band **4** _{3}, and HH sub-band **4** _{4}) shown in **2**LL to **1**HH in total. In

In the above description, in the case of performing a wavelet transform only one-dimensionally, the wavelet transform may be performed in either a horizontal or a vertical direction.

On the other hand, inverse wavelet transform is performed by first performing inverse low-pass filtering horizontally on each coefficient of an even X-coordinate value x (x=2i+1) using its adjacent coefficients and then performing inverse high-pass filtering on each coefficient of an odd X-coordinate value x (x=2i) using its adjacent results of inverse low-pass filtering in the interleaved coefficient array **3** shown in

*P*(2*i*)=*C*(2*i*)−└(*C*(2*i−*1)+*C*(2*i+*1)+2)/4┘ (3)

*P*(2*i+*1)=*C*(2*i+*1)+└(*P*(2*i*)+*P*(2*i+*2))/4┘ (4)

As in the above-described case of a wavelet transform, in the edge units of the image, a pixel in the center may lack an adjacent pixel. In this case, an appropriate pixel value is supplemented in accordance with a predetermined rule, which is irrelevant to the essence of the present invention so that a description thereof will be omitted.

In this manner, the coefficient array **3** of **2**; that is, the coefficient array **3** is inversely converted. Likewise, successively thereafter, inverse low-pass filtering is performed vertically on each coefficient of an even Y-coordinate value y (y=2i) using its adjacent coefficients, and then, inverse high-pass filtering is performed on each coefficient of an odd Y-coordinate value (y=2i+1) using its adjacent results of inverse low-pass filtering. This process is performed on each X-coordinate value x. In so doing, one inverse wavelet transform operation is completed, so that the coefficient array **2** of **1** of the image of **1** of the original image is restructured. When a plurality of wavelet transform operations are performed, the same inverse wavelet transform operation may be repeated using the HL and other coefficients, considering the coefficient array **1** of

At this point, it can be understood that the result of inverse low-pass filtering is obtained from three coefficients including a coefficient from the filtering center (that is, a coefficient desired to be subjected to filtering), and it can also be understood that the result of inverse high-pass filtering is obtained from a coefficient from the filtering center and the two adjacent results of inverse low-pass filtering. The two results of inverse low-pass filtering are calculated from five coefficients including the coefficient from the filtering center. Therefore, the result of inverse high-pass filtering is obtained from five coefficients. As is apparent from the above-described understanding, in order to obtain one pixel, it is sufficient to perform an inverse wavelet transform centered on the target pixel, and it is not necessary to perform inverse wavelet transform on pixels irrelevant to the inverse wavelet transform on the target pixel. Accordingly, by performing an inverse wavelet transform using only those required to finally obtain a desired pixel of all of the coefficients in the same sub-band, a decoded image of a desired resolution can be generated with a reduced amount of processing without obtaining pixels to be subtracted.

**10** according to a first embodiment of the present invention.

The image data generator **10** of this embodiment includes an inverse wavelet transform unit **11** and a coefficient selecting unit **12**. The inverse wavelet transform unit **11** performs inverse wavelet transform on wavelet coefficients obtained by performing wavelet transform on an original image one or a plurality of times, so that the image data generator **10** generates data on a reduced size image of the original image. The original image includes an image obtained after performing conversion such as color conversion on the original image.

The coefficient selecting unit **12** selects those to be subjected to inverse wavelet transform of the coefficients in the same sub-band. In other words, all of the coefficients before selection can also be defined as all of the coefficients in one LL sub-band, which may be data on the original image. That is, the LL sub-band may be four sub-bands having a decomposition level higher than that of the LL sub-band by one. The selecting operation in this embodiment is different from the conventional technique of, for instance, selecting **2**LL, **2**HL, **2**LH, and **2**HH sub-bands but not selecting (that is, not performing inverse wavelet transform on) **1**HL, **1**LH, and **1**HH sub-bands. Practically, as will be later described, the selecting operation can be performed more simply if the coefficients before selection are interleaved.

The inverse wavelet transform unit **11** performs inverse wavelet transform only on the coefficients selected by the coefficient selecting unit **12** from the above-described wavelet coefficients in the same sub-band. In this manner, data on the reduced size image is generated. For instance, in the case of performing an inverse wavelet transform on the coefficients selected from the coefficients in the **2**LL sub-band (that is, **3**LL, **3**HL, **3**LH, and **3**HH sub-bands) using the selective inverse wavelet transform of one embodiment of the present invention, the wavelet coefficients relating to other operations including an operation preceding the inverse wavelet transform, such as coefficients in the **3**LL sub-band (composed of **4**LL, **4**HL, **4**LH, and **4**HH sub-bands and not processed through the coefficient selecting unit **12**) to be obtained as required in the preceding operation may be inversely converted by the inverse wavelet transform unit **11**.

As described above, it may also be considered that the image data generator **10** according to this embodiment includes a unit to selectively subject coefficients in the same or a single sub-band to inverse wavelet transform. Here, the sub-band may be the entire original image.

In the specification, coefficients to be inversely converted (or subjected to inverse wavelet transform) by the inverse wavelet transform unit **11** are referred to as wavelet coefficients or simply as coefficients. Coefficients to be again inversely converted after being subjected to an inverse wavelet transform are also referred to as wavelet coefficients or simply as coefficients. In the case of two-dimensional inverse wavelet transform, coefficients to be subjected to an inverse wavelet transform in both directions and coefficients to be subjected to an inverse wavelet transform in one direction are also referred to as wavelet coefficients or simply as coefficients.

For purposes of this disclosure, the size of the original image is defined as 16×16, and a 1/3 scale image, or an image reduced to one third of the original image vertically and horizontally in size, is generated from the original image as previously described in the first prior art and the second prior art. For purposes of this disclosure, the 1/3 scale image is a 5×5 image since **16** is not divisible by three. **1** of

Additionally, in order to generate a 5×4 image from the 16×16 image, an inverse wavelet transform may be performed centered only on the coefficients of the underlined pixels of all of the coefficients in

The coefficient selecting unit **12** may include a unit to select at least all of the wavelet coefficients relating to the pixels of a reduced size image to be generated (or desired to be finally obtained). Further, the coefficient selecting unit **12** may include a unit to select only the wavelet coefficients relating to the pixels of a reduced size image to be generated (or desired to be finally obtained). That is, the inverse wavelet transform may be performed only on the wavelet coefficients relating to the pixels of a reduced size image desired to be obtained. For purposes of this disclosure, the term “relating” refers to a relationship originating from the nature of wavelet transforms, such as a relationship derived from the above-described equations (1) through (4) (the shaded pixels to the underlined pixels in

It should be noted that an image obtained as a result of performing an inverse wavelet transform only on the coefficients relating to the pixels of a reduced size image desired to be obtained is mathematically equivalent to a reduced size image generated by thinning out or reducing an image obtained by performing an inverse wavelet transform on all of the wavelet coefficients. In the case of thinning out the wavelet coefficients themselves and performing an inverse wavelet transform on the remaining wavelet coefficients, a reduced size image of the same size can be obtained. In this case, however, the inverse wavelet transform is performed using coefficients different from those obtained in the “forward” wavelet transform. Therefore, the obtained reduced size image has no mathematical grounds.

On the other hand, according to one embodiment of the present invention, a totally equivalent image having mathematical grounds can be obtained.

As previously described, in order to obtain the underlined 25 pixels of

Accordingly, all of the coefficients positioned on a straight line extending in one direction (in which the inverse wavelet transform is performed in an inverse wavelet transform operation) where the straight line includes a coefficient corresponding to a pixel position of a reduced size image desired to be generated may be obtained. The coefficient selecting unit **12** may include a unit to select, of the above-described wavelet coefficients in the same sub-band (in the interleaved state), all of the coefficients positioned on a straight line extending in one direction (in the vertical direction in the above-described case) in which an inverse wavelet transform is performed, the straight line including a wavelet coefficient in a position of a pixel of a reduced size image desired to be generated. By this selecting operation, all of these selected wavelet coefficients are subjected to an inverse wavelet transform in the inverse wavelet transform unit **11**, and as a result of this inverse wavelet transform, all of the coefficients positioned on each of the above-described straight lines including the coefficients of the pixel positions of the reduced size image desired to be generated. The pixels composed of the obtained coefficients are thinned out as appropriate as described above, so that the desired reduced size image is obtained. According to this configuration, the actual application can be simplified in generating a reduced size image.

**7** of **2** of

In order to obtain the coefficients of the shaded pixel positions of

Accordingly, an inverse wavelet transform may be performed two-dimensionally, and all of the coefficients positioned on straight lines perpendicular to the direction in which the first inverse wavelet transform in an inverse wavelet transform operation is performed may be obtained, where the straight lines include the coefficients corresponding to the positions of the pixels of the reduced size image and the positions relating to the positions of the pixels. The coefficient selecting unit **12** of this embodiment may include a unit that, at the time of selecting coefficients for inverse wavelet transform performed in one direction (in the horizontal direction in the above-described case) in which the inverse wavelet transform is performed first in an inverse wavelet transform operation, selects, of the above-described wavelet coefficients in the same sub-band (in the interleaved state), all of the coefficients on straight lines perpendicular to the direction in which the inverse wavelet transform is performed first, where the straight lines include the coefficients corresponding to the positions of the pixels of the reduced size image to be generated and the positions of pixels necessary for obtaining the positions of the pixels of the reduced size image. As a result of employing this unit of the coefficient selecting unit **12**, the selected coefficients are subjected to an inverse wavelet transform in the direction in which the inverse wavelet transform is performed first, and are subjected to an inverse wavelet transform in the other direction and thinned out as required. In so doing, the coefficients of the pixel positions of the reduced size image desired to be generated are obtained. According to this configuration, the actual application can be simplified in generating a reduced size image.

In **2**LL coefficients of a one-fourth resolution, may be obtained by performing an inverse wavelet transform. Then, the obtained **2**LL coefficients are selectively subjected to inverse wavelet transform using the coefficients of the **2**HL, **2**LH, and **2**HH sub-bands, so that a 1/3-resolution image can be finally obtained.

Therefore, according to one embodiment of the present invention, an inverse wavelet transform may be performed on all of the coefficients of up to the resolution closest to a desired resolution. That is, the coefficient selecting unit **12** of one embodiment of the present invention may include a unit that selects all of the wavelet coefficients in the sub-bands of the same decomposition level higher than that of the LL sub-band having the highest resolution below the resolution of a reduced size image to be obtained so that the LL sub-band is obtained by an inverse wavelet transform.

As shown in the shaded units of

Therefore, according to one embodiment of the present invention, the coefficients of pixels in such positions as to reduce the sum of the tap lengths of the filters may be selected. In

In the case of performing inverse wavelet transform two-dimensionally, the operation of performing an inverse wavelet transform selectively on the coefficients in the same (or a single) sub-band, and the operation of causing the sum of the tap lengths of the high-pass and low-pass filters to be smaller than n(Tlow+Thigh), the high-pass and low-pass filters having different tap lengths for an inverse wavelet transform and performing filtering centered on the coefficients corresponding to the pixel positions of a reduced size image of all of the coefficients in the interleaved state may be provided. In the case of generating a reduced size image by performing the inverse wavelet transform two-dimensionally, processing can also be performed at high speed.

In the case of generating the 1/3-resolution reduced size image of an original image from wavelet coefficients divided into **3**LL, **3**HL, **3**LH, **3**HH, **2**HL, **2**LH, **2**HH, **1**HL, **1**LH, and **1**HH sub-bands, using the above-described configurations suitably, the following processes (a) through (c) may be performed as main procedures to which the selective inverse wavelet transform of one embodiment of the present invention is applied. In the case of performing an inverse wavelet transform on two-dimensional coefficients in the horizontal and vertical directions, it may be determined based on the direction in which the inverse wavelet transform is performed whether the selective inverse wavelet transform of one embodiment of the present invention is employed.

(a) The conventional inverse wavelet transform, which does not employ the selective inverse wavelet transform of one embodiment of the present invention, that is, which selects all of the coefficients, is performed on the coefficients of the **2**LL sub-band, that is, the coefficients of the four sub-bands of decomposition level **3** (**3**LL, **3**HL, **3**LH, and **3**HH). By doing so, the **2**LL sub-band is obtained. Then, the coefficients of the **1**LL sub-band, that is, the coefficients of the four sub-bands of decomposition level **2** (**2**LL, **2**HL, **2**LH, and **2**HH) in the interleaved state, are selectively subjected to the inverse wavelet transform, so that a new LL sub-band is extracted. In this manner, a reduced size image is obtained.

(b) Coefficients are selected, in consideration of two inverse wavelet transform operations, from the coefficients of the **2**LL sub-band, that is, the coefficients of the four sub-bands of decomposition level **3** (**3**LL, **3**HL, **3**LH, and **3**HH), in the interleaved state, and the selected coefficients are subjected to an inverse wavelet transform twice using the coefficients of the **2**HL, **2**LH, and **2**HH sub-bands. In this manner, a reduced size image is obtained.

(c) The selection of coefficients in each of the processes (a) and (b) is simplified, and the unnecessary coefficients after the inverse wavelet transform operations are subtracted so that a reduced size image is obtained.

In the case of compressing an original image, which is, for instance, a color image formed of three components of R, G, and B, normally, in order to increase the compression rate, color conversion is performed on the R, G, and B components so that the R, G, and B components are converted to three new components of one luminance and two color differences, and a wavelet transform is performed on each of the new components obtained after the color conversion.

A color conversion called RCT (Reversible Component Transform), which is one of color conversion methods employed in JPEG 2000, is given by the following:

The inverse conversion of RCT is given by the following:

JPEG 2000 is an image compression and decompression method standardized internationally in 2001 as a successor to JPEG. The image data generator of **22**, a quantization and inverse quantization unit **23**, and an entropy coding and decoding unit **24**. Of these units, the two-dimensional wavelet transform and inverse transform unit **22** relates to one embodiment of the present invention. In the specification, the two-dimensional wavelet transform and inverse transform unit **22** is described mainly as an inverse wavelet transform unit. The two-dimensional wavelet transform and inverse transform unit **22**, the quantization and inverse quantization unit **23**, and the entropy coding and decoding unit **24** are provided as a unit receive an input from and transmit an output to a color space conversion and inverse conversion unit **21** and a tag processing unit **25**. Each of these units may have independent configurations for the forward (compression) and inverse (decompression) operations. Further, each of these units may perform processing component by component. At the time of performing decompression in the JPEG 2000 configuration of

When the effect of each component on image quality is considered, normally, the color-difference components are more quantized in the quantization process of

Therefore, according to this embodiment, an apparatus for generating a reduced size image of an original image by performing an inverse wavelet transform on the wavelet coefficients of a plurality of components may include a unit that performs the inverse wavelet transform on the components to different resolutions so that the resolution differs between the components when the inverse wavelet transform is performed. That is, an inverse wavelet transform may be performed with a difference between the resolutions of the components. The image data generator of this embodiment may be an apparatus for generating data on a reduced size image of an original image, the apparatus including an inverse wavelet transform unit that performs an inverse wavelet transform on the wavelet components of a plurality of components. The inverse wavelet transform unit performs an inverse wavelet transform component by component with the resolution differing between at least one of the components and the rest of the components. According to this embodiment, a reduced size image having a plurality of components can be generated at high speed with a reduced amount of processing using the characteristic between the components.

Moreover, by applying the characteristics of this embodiment to the above-described first embodiment of the image data generator relating to the selective inverse wavelet transform, processing speed can be further increased and the actual application can be further simplified.

Further, with respect to each of the wavelet coefficient luminance component and the wavelet coefficient color-difference component of the above-described plural components, the coefficients of the LL sub-band having the highest resolution below the resolution of a reduced size image to be generated may be obtained. According to this configuration, a reduced size image formed of a plurality of components can be generated at higher speed with more simplicity.

Further, in the image data generator of this embodiment, the above-described components may be a luminance component and a color-difference component, and the luminance component is subjected to an inverse wavelet transform to a resolution higher than a resolution to which the color-difference component is subjected to an inverse wavelet transform. As a result, the color-difference component has a lower resolution. According to this configuration, a reduced size image formed of luminance and color-difference components can be generated at higher speed with a reduced amount of processing while the effect on image quality is controlled.

Further, according to this embodiment, the above-described components may include two color-difference components, and the two color-difference components may be subjected to an inverse wavelet transform to different resolutions. According to this configuration, a reduced size image formed of a plurality of components may be generated at higher speed by setting a lower resolution for one of the color-difference components than for the other one of the color-difference components, using the characteristic between the color-difference components.

Further, in this embodiment, the resolution of the color-difference components may be set to a half of the resolution of the luminance component. Generally, this resolution combination, which corresponds to sub-sampling color difference to one-fourth of luminance, provides a good balance between image quality and processing speed. According to this configuration, a reduced size image having a plurality of components can be generated with image quality and processing speed being in balance with each other.

The foregoing embodiments are described, focusing on the image data generators according to the present invention. Embodiments of the present invention can also be realized as an image data generating method including the processing operations performed in the image data generators; a program for causing a computer to function as any of the above-described image data generators or as each unit thereof or a program for causing a computer to execute the image data generating method (a computer program with the contents of the processing steps); and a computer-readable recording medium recording the program (a computer-readable information recording medium recording the contents of the processing steps). As is apparent from a later-described embodiment, this image data generating method of the present invention may be used to generate a reduced size image with a reduced amount of calculation and/or generate a reduced size image at high speed using the characteristics between components. Further, one embodiment of this program makes it possible to provide a system that can generate a reduced size image with a reduced amount of calculation, can generate a reduced size image simply with high image quality, and can generate a reduced size image at high speed using the characteristics between components by the operations corresponding to the above-described embodiments. Further, one embodiment of this recording medium makes it possible to provide a system that can generate a reduced size image with a reduced amount of calculation, can generate a reduced size image simply with high image quality, and can generate a reduced size image at high speed using the characteristics between components with the configurations corresponding to the above-described embodiments. It is apparent that this program and this recording medium can be realized easily based on the later-described embodiment in addition to the above-described embodiments.

A description will be given of a recording medium storing a program or data for realizing the image data generating function of the present invention according to a third embodiment of the present invention.

Specifically, the recording medium may be any of a CD-ROM, a magneto-optical disk, a DVD-ROM, a FD, a flash memory, and other various ROMs and RAMs. The recording medium is recorded with a program for causing a computer to execute the functions relating to the above-described embodiments according to the present invention so as to realize the image data generating function of the present invention. By distributing this recording medium, the above-described functions can be realized easily. Then, the image data generating function according to one embodiment, the present invention can be performed by loading the recording medium into an information processing apparatus such as a computer and causing the information processing apparatus to read out the program or by storing the program in a recording medium provided in an information processing apparatus and reading out the program as required.

An expatiation will be given of the third embodiment of the present invention.

According to the image data generator of **31**, a CPU **32**, and an HDD **34** are connected via a data bus **33**. In one embodiment, a reduced size image is generated from a compressed image of an original resolution and stored in the HDD **34** using the following process.

In step S**201**, compressed image data of an original resolution recorded on the HDD **34** is read into the RAM **31** by a command transmitted from the CPU **32**. Next, in step S**202**, the CPU **32** reads the compressed image data recorded in the RAM **31** only for an amount corresponding to a predetermined resolution, and obtains wavelet coefficients. The CPU **32** performs processing on the wavelet coefficients according to one embodiment of the present invention, thereby generating a reduced size image. In step S**203**, the CPU **32** writes the generated reduced size image to another region in the RAM **31**. In step S**204**, the reduced size image is recorded on the HDD **34** by a command transmitted from the CPU **32**.

According to the image data generator of **35** inside a personal computer (PC), a CPU **36** inside the PC, an HDD **38**, and a monitor **39** such as a CRT or LCD monitor are connected via a data bus **39**. By an instruction from a user, a reduced size image is generated at high speed from image data of an original resolution and displayed on the monitor **39**.

In step S**211**, compressed image data of an original resolution recorded on the HDD **38** is read into the RAM **35** by a command transmitted from the CPU **36**. Next, in step S**212**, the CPU **36** reads the compressed image data recorded on the RAM **35** only for an amount corresponding to a predetermined resolution, and obtains wavelet coefficients. Then, the CPU **36** performs processing on the wavelet coefficients according to one embodiment of the present invention, thereby generating a reduced size image. In step S**213**, the CPU **36** writes the generated reduced size image to another region in the RAM **35**. In step S**214**, the reduced size image is transmitted to and displayed on the monitor **39** by a command transmitted from the CPU **36**.

A description will be given of the case where a reduced size image is generated based on an image defined by the even number of rows of pixels and the even number of columns of pixels, where the image is compressed in compliance with a JPEG-2000 format using the color conversion of the equations (5) and 5×3 wavelet transform in an image data generator having the above-described configuration according to one embodiment of the present invention.

In the operation of generating a reduced size data in this embodiment, first, in step S**1**, the resolution level of a reduced size image to be generated is set to (resolution level) r based on a resolution (1/k) specified by a user. It is preferable that the image data generator include a unit that allows the user to specify a desired resolution. Then, in step S**2**, (resolution level) i is set to zero. In steps S**3** through S**5**, i is incremented until i reaches r. During this process, in step S**4**, the entropy codes of all components are decoded with respect to each resolution level i. The decoding of the entropy codes is the decoding of variable length codes, and an MQ-coder is employed for the decoding in JPEG 2000. Since the decoding of the entropy codes is irrelevant to the point of this embodiment, a description thereof will be omitted.

**0** (luminance), component **1** (color difference Cb), and component **2** (color difference Cr) of resolution **0** (minimum resolution), components **0**, **1**, and **2** of resolution **1**, components **0**, **1**, and **2** of resolution **2**, . . . . As is well known, in JPEG 2000, the codes of all components can be arranged in the order of resolutions as shown in **1** through S**5**).

In this specification, the expression of a 1/3 “resolution” is frequently used, while in JPEG 2000, a quantity called “resolution level” is employed as a quantity similar to “resolution.” The lowest resolution is 0, and the second lowest resolution is 1. **3**LL is resolution level **0**, the resolution level of **3**HL, **3**LH, and **3**HH is resolution level **1**, the resolution level of **2**HL, **2**LH, and **2**HH is resolution level **2**, and the resolution level of **1**HL, **1**LH, and **1**HH is resolution level **3**. Further, the resolution level of **2**LL is resolution level **1**, and the resolution level of **1**LL is resolution level **2**.

The ratio of the resolution of revolution level r to the resolution of an original image depends on the maximum value of the decomposition level. If the maximum value of the decomposition level is d, the resolution of the wavelet coefficients (sub-band) of resolution level r (r>0) with respect to the original image is 1/2^{(d−r+1)}. Therefore, the value of r that should be set in step S**1** in order to generate a 1/k scale image of the original image is given by:

*r=d−(log* _{2} *k−*1) (7)

Accordingly, when the desired resolution (1/k) with respect to the original image is provided by the user, the value of r is calculated by the equation (7) to be set in step S**1**. In the case of generating a reduced size image having an aspect ratio different from that of its original image as shown in **1**, so that the operation of

Normally, the wavelet coefficients are quantized, and therefore, should be inversely quantized before being subjected to an inverse wavelet transform. Therefore, when the decoding is completed in each resolution level i lower than resolution level r (that is, “NO” in step S**3**), in step S**6**, all of the decoded wavelet coefficients are inversely quantized. Thereafter, in step S**7**, the inversely quantized wavelet coefficients are subjected to a selected inverse wavelet transform operation. The selected inverse wavelet transform operation of step S**7** is composed of an LL generating operation performed shown in **19**, and **21** according to this embodiment is performed with respect to each component. In step S**8**, an inverse color conversion operation is performed when there are two or more components.

According to this embodiment, first, in step S**11** of **12**, r is decremented by two (r=r−2), and the LL generating operation is performed on the color difference Cb. Then, in step S**13**, r is incremented by one (r=r+1), and the LL generating operation is performed on the color difference Cr. Similarly, when the resolution of luminance is reduced in the ratio of k:1, the resolution of Cr and the resolution of Cb are reduced in the ratio of 2k:1 and 4k:1, respectively. Then, as far as Cr and Cb are concerned, k in

**14** of **15** through S**17**, i is incremented until i reaches r. During this process, in step S**16**, with respect to each resolution level i, using an LL sub-band and three HL, LH, and HH sub-bands of resolution level i, inverse wavelet transform is performed to generate a new LL sub-band for a specified component. When the generation of a new LL sub-band is completed in each resolution level i below resolution level r (that is, “NO” in step S**15**), in step S**18**, the number of vertical pixels (rows) LLheight and the number of horizontal pixels (columns) LLwidth of the new LL sub-band are retained. Then, in step S**19**, the LL generating operation ends. In this manner, all of the wavelet coefficients of the LL sub-band having the highest resolution below the resolution of a reduced size image to be generated are obtained.

As described in step S**12**, with respect to the color difference Cb, the LL sub-band is generated using the sub-bands of up to a resolution level lower than that of the luminance Y by two. With respect to the color difference Cr, the LL sub-band is generated using the sub-bands of up to a resolution level higher than that of the color difference Cb by one. In the case of

According to this embodiment, first, in step S**21** of **22**, the horizontal pixel position determining and horizontal filtering operation is performed on the color difference Cb. Then, in step S**23**, the horizontal pixel position determining and horizontal filtering operation is performed on the color difference Cr.

^{n }smaller than k, obtaining the horizontal positions of the above-described “pixels desired to be finally obtained” with k/p being employed as a unit (a horizontal position determining operation), and performing horizontal filtering on all of the vertical positions of each of the horizontal positions. Although horizontal filtering is performed on all of the vertical positions, the wavelet coefficients can be selected so as to be selectively subjected to an inverse wavelet transform. This is because the horizontal position determining operation is performed. That is, according to the horizontal pixel position determining and horizontal filtering operation, first, in step S**24** of ^{n }smaller than k is obtained based on the resolution 1/k specified by the user. Next, in step S**25** i is set to zero (i=0). Then, in step S**26**, a horizontal position xs is determined by xs=└ik/p┘. Thereafter, in steps S**27** and S**28**, i is incremented until xs equals 2×LLwidth (xs=2×LLwidth), while step S**26** is performed. During this process, in step S**29**, horizontal filtering is performed on all of the vertical positions of each column whose X-coordinate value is xs. If the result of step S**27** is “NO”, the operation ends. As stated above, when the aspect ratio of a reduced size image is different from that of the original image, if a desired resolution (1/kh horizontally and 1/kv vertically) with respect to the original image is provided by a user, the value of r is calculated by the equation (7) using the smaller of kh and kv as k.

At this point, the number of adjacent pixels to be obtained can be reduced by first selecting (**0**, **0**) of

According to this embodiment, first, in step S**31**, the vertical pixel position determining and vertical filtering operation is performed on the luminance Y. Next, in step S**32**, the vertical pixel position determining and vertical filtering operation is performed on the color difference Cb. Then, in step S**33**, the vertical pixel position determining and vertical filtering operation is performed on the color difference Cr.

The vertical pixel position determining and vertical filtering operation, which is similar to the horizontal pixel position determining and horizontal filtering operation of **41**, i and j are set to zero (i=j=0). Then, in step S**42**, a horizontal position xs is determined by xs=└ik/p┘. In step S**43**, it is determined whether xs<2×LLwidth holds. If xs<2×LLwidth (that is, “YES” in step S**43**), steps S**44** and S**45** (will be later described) are performed. In the case of “NO” in step S**45**, in step S**46**, i is incremented by one and j is set to zero (j=0). If xs≧2×LLwidth (that is, “NO” in step S**43**), the operation ends. In step S**44**, a vertical position ys is determined by ys=└jk/p┘. In step S**45**, it is determined whether ys<2×LLheight holds. If ys<2×LLheight (that is, “YES” in step S**45**), steps S**48** and S**49** (will be later described) are performed, and in step S**47**, j is incremented by one (j=j+1), and the operation returns to step S**44**. This process continues until ys equals 2×LLheight. If ys≧2×LLheight (that is, “NO” in step S**45**), in step S**46**, i is incremented by one and j is set to zero (j=0), and the operation returns to step S**42**. In step S**48**, vertical filtering is performed only on each pixel position whose Y-coordinate value is ys. In step S**49**, each pixel value after the filtering is stored in the position (i, j) in a buffer for reduced component. Thereafter, as previously described, step S**47** is performed. In **42** and the vertical position ys is determined by ys=└jkv/p┘ in step S**44**.

The maximum value of the decomposition level and the number of vertical pixels (height) and the number of horizontal pixels (width) of an original image can be read from the headers of the codes, and the number of vertical pixels and the number of horizontal pixels of a reduced size image to be generated are given by └height/k┘ and └width/k┘, respectively, and also used in the inverse color conversion operation. The color difference components are subjected to an inverse wavelet transform only up to resolutions lower than that of the luminance component by one or more levels. Therefore, the number of pixels of the reduced size image to be generated is defined by the following:

Color difference Cb:

Vertical pixels └height/4k┘,

Horizontal pixels └width/4k┘

Color difference Cr:

Vertical pixels └height/2k┘,

Horizontal pixels └width/2k┘

Similarly, when the aspect ratio of a reduced size image is different from the original image, the number of pixels of the reduced size image to be generated is defined by the following:

Color difference Cb:

Vertical pixels └height/4kv┘,

Horizontal pixels └width/4kh┘;

Color difference Cr:

Vertical pixels └height/2kv┘,

Horizontal pixels └width/2kh┘

In the inverse color conversion operation, if i<└height/k┘ and j<└width/k┘, later-described steps S**55** through S**59** are performed. That is, first, in step S**51** of **52**) in step S**54**, i is incremented by one until i equals └height/k┘ with j being set to zero (j=0). During this operation, below-described operations are performed, and if it is determined in step S**52** that i≧└height/k┘, the operation ends. The operations performed during the above-described operation are as follows. In the case of “YES” in step S**53**, steps S**55** through S**59** are performed until j equals └width/k┘. If it is determined in step S**53** that j≧└width/k┘ (that is, “NO” in step S**53**), i is incremented and j is set to zero (j=0), and the operation returns to step S**52**. In step S**55**, with respect to the luminance Y, Y is determined to be the value of the position (i, j) in the buffer for reduced component of the luminance Y. In step S**56**, with respect to the color difference Cb, Cb is determined to be the value of the position (i/**4**, j/**4**) in the buffer for reduced component of the color difference Cb. In step S**57**, with respect to the color difference Cr, Cr is determined to be the value of the position (i/**2**, j/**2**) in the buffer for reduced component of the color difference Cr. Next, in step S**58**, an inverse color conversion is performed by the above-described equations (6), and in step S**59**, the R, G, and B values are stored in the position (i, j) in a buffer for reduced size image.

As previously described, in the case of obtaining the 1/3-resolution image of an original image, the wavelet coefficients of up to an LL sub-band (and HL, LH, and HH sub-bands) that may be interleaved to have a resolution (or pixels) half the resolution (or pixels) of the original image are obtained, and the 2/3-resolution image of the interleaved coefficients of the LL sub-band is generated by the processing described herein.

**8** of **9** of

The above description is given of the selective inverse wavelet transform operation according to one embodiment of the present invention. A description will now be given of an image data generator, an image data generating method, and an image data generating program for generating decoded image data of a desired resolution (desired reduced size image data), balancing decoding speed and decoded image quality by decoding unit of the codes of wavelet coefficients in an entropy decoding unit in generating a reduced size image. A description will also be given of a recording medium storing such a program.

Decoding is often employed in generating a reduced size image to be displayed on a display unit having a small screen. In this case, since the display unit is limited in representation performance, the image quality of the reduced size image itself may not be strictly questioned. Rather, decoding time may become an issue. Further, due to the limited representation performance of the display unit, the degradation of data on the reduced size image often cannot be observed visually.

In a method that can make a selection as to not only resolution (band) but also the ratio of codes to be decoded (code-decoding ratio) in an entropy decoding unit, such as JPEG 2000, the entropy decoding of wavelet coefficients can be performed at high speed by limiting the code-decoding ratio. In the case of JPEG 2000, since the binary representation of a wavelet coefficient is decoded bit by bit from MSB (Most Significant Bit) to LSB (Least Significant Bit), decoding image quality and decoding speed can be improved by selecting an appropriate bit up to which the decoding is to be performed. In the case of displaying an image on a display unit having a limited display capability as previously described, it is often the case that it is not necessary to decode all codes.

**40** according to this embodiment.

The image data generator **40**, which generates data on a reduced size image of an original image, includes a coefficient decoding unit **41** decode coded wavelet coefficients and a code omitting unit **42**.

The code omitting unit **42** omits, of the codes of the same sub-band, the codes that are not to be decoded. The coefficient decoding unit **41** decodes the codes other than those omitted in the code omitting unit **42**. That is, the coefficient decoding unit **41** decodes unit of the codes of the same sub-band. As will be later described, the image data generator **40** may decode unit of a layer and obtain a reduced size image based on the decoded data. According to this embodiment, it is possible to generate reduced size data, balancing decoding speed and decoded image quality. The partial decoding according to one embodiment of the present invention excludes the case of performing no decoding, which case is equivalent to the case of omitting decoding by band restriction.

In realizing the above-described improvements, it is important to reduce the amount of codes to be decoded while controlling the effect on image quality. It is well known that in the case of a wavelet transform, a quantization error generated in coefficients affects image quality differently in each sub-band (each of LL, HL, LH, and HH). The LL sub-band has the largest effect on image quality, and generally, HL≈LH>HH holds in terms of the magnitude of the effect on image quality. Accordingly, normally, the LL sub-band is least quantized while the HH sub-band is most quantized. Further, as in the above-described case of JPEG 2000, decoding only a number of high-order bit planes (will be later described) with the remaining low-order bit planes being left undecoded is equivalent to losing lower bits by quantization. This means that it is possible to provide a greater code-decoding ratio to the LL sub-band and a smaller code-decoding ratio to the HH sub-band.

Further, it is well know that in the case of a wavelet transform, the effect of quantization error generated in coefficients on image quality differs depending on the decomposition level. The effect becomes greater as the decomposition level becomes higher.

Therefore, according to this embodiment, the sub-bands may be provided with different code-decoding ratios. That is, the amount or ratio of codes to be omitted (not to be decoded) in the code omitting unit **42** may be caused to differ among the sub-bands. According to this configuration, by considering the effect on image quality by each sub-band, a reduced size image can be generated with decoding speed and decoded image quality being improved.

Further, according to this embodiment, the LL, HL, LH, and HH sub-bands may be provided with different code-decoding ratios. That is, the amount or ratio of codes to be omitted in the code omitting unit **42** may be caused to differ among the sub-bands of the same decomposition level. According to this configuration, by considering the effect on image quality by each of the sub-bands of the same decomposition level, a reduced size image can be generated with decoding speed and decoded image quality beingimproved improved.

Further, according to this embodiment, the decomposition levels may be provided with different code-decoding ratios. That is, the amount or ratio of codes to be omitted in the code omitting unit **42** may be caused to differ among the decomposition levels. According to this configuration, by considering the effect on image quality by each decomposition level, a reduced size image can be generated with decoding speed and decoded image quality being improved.

In the case of compressing an original image, which is, for instance, a color image formed of three components of R, G, and B, normally, in order to increase the compression rate, a color conversion is performed on the R, G, and B components so that the R, G, and B components are converted to three new components of one luminance and two color differences, and wavelet transform is performed independently on each of the new components obtained after the color conversion.

As previously described, RCT, which is one of color conversion methods employed in JPEG 2000, is given by the following:

The inverse conversion of RCT is given by the following:

where the symbol └x┘ represents a floor function of x that replaces a real number x with the largest of all the integers smaller than or equal to x.

In this case, it is well known that the effect of a quantization error generated in coefficients on image quality differs among the components. In the case of the luminance-color difference system, the luminance exerts the largest effect. Further, it is also well known that the effect differs between the color differences. In the case of no color conversion, the G component has the largest effect.

Therefore, according to one embodiment of the present invention, in an apparatus decoding the wavelet coefficients of a plurality of components, the components may be provided with different code-decoding ratios. That is, decoding may be performed component by component, and the amount or ratio of codes to be omitted in the code omitting unit **42** may be caused to differ among the components. According to this configuration, by considering the effect on image quality by each component, a reduced size image can be generated with decoding speed and decoded image quality being improved.

Further, according to this embodiment, the LL sub-band may be provided with the smallest amount or ratio of codes not to be decoded. According to this configuration, by considering the effect on image quality by the LL sub-band, a reduced size image can be generated with decoding speed and decoded image quality being improved.

Further, according to this embodiment, the HH sub-band may be provided with the largest amount or ratio of codes not to be decoded. According to this configuration, by considering the effect on image quality by the HH sub-band, a reduced size image can be generated with decoding speed and decoded image quality being improved.

Further, according to this embodiment, a higher decomposition level may be provided with a smaller amount or ratio of codes not to be decoded. According to this configuration, by considering the effect on image quality by the decomposition level, a reduced size image can be generated with decoding speed and decoded image quality being improved.

Further, according to this embodiment, in an apparatus decoding a luminance component, the luminance component may be provided with the smallest amount or ratio of codes not to be decoded of all components. According to this configuration, by considering the effect on image quality by luminance, a reduced size image can be generated with decoding speed and decoded image quality being improved.

Further, according to this embodiment, in an apparatus decoding a plurality of color-difference components, the color-difference components may be provided with different amounts or ratios of codes not to be decoded. According to this configuration, by considering the effect on image quality by each color-difference component, a reduced size image can be generated with decoding speed and decoded image quality being improved.

Further, according to this embodiment, in an apparatus decoding a luminance component Y and two color-difference components Cb and Cr, Y<Cr<Cb may hold in terms of the amount or ratio of codes not to be decoded. It is empirically known that in light of human visual characteristics, Cb has a smaller effect than Cr. According to this configuration, by considering the effect on image quality by each of Cb and Cr, a reduced size image can be generated with decoding speed and decoded image quality being improved.

Normally, in order to generate a reduced size image of a desired size, codes are decoded to a size larger than the desired size and thinned out after being subjected to an inverse wavelet transform, or codes are decoded to a size smaller than the desired size and subjected to interpolation. When these decoding cases are compared at the same decomposition level (to be subjected to decoding), the code-decoding ratio should be set higher in the latter case (with interpolation processing) than in the former case (with thinning-out processing).

Therefore, the image data generator **40** of this embodiment may include a unit to decode only a sub-band of a size smaller than or equal to a desired image size. In other words, the codes are extracted in the size of an LL sub-band. According to this configuration, by restricting the size of a sub-band to be decoded, a reduced size image can be generated with decoding speed and decoded image quality being improved.

Further, the image data generator **40** of this embodiment may include a unit to decode only a sub-band of a size smaller than a desired image size and a unit to thin out the decoded sub-band. According to this configuration, by restricting the size of a sub-band to be decoded, a reduced size image can be generated with decoding speed and decoded image quality being improved.

Further, the image data generator **40** of this embodiment may include a unit to decode sub-bands of up to the maximum of sizes smaller than or equal to a desired image size and a unit to perform interpolation. According to this configuration, by restricting the size of a sub-band to be decoded, a reduced size image can be generated with decoding speed and decoded image quality being improved.

The above description is given, focusing on the image data generator performing the partial decoding of one embodiment of the present invention. One embodiment of the present invention may also be realized as: an image data generating method including the processing steps performed in the image data generator; a program for causing a computer to function as the above-described image data generator or as each unit thereof or a program for causing a computer to execute the image data generating method (a computer program with the contents of the processing steps); and a computer-readable recording medium recording the program (a computer-readable information recording medium recording the contents of the processing steps). As is apparent from a later-described embodiment, by this image data generating method of the present invention, a reduced size image can be generated with decoding speed and decoded image quality being balanced or improved. Further, this program makes it possible to provide a system that can generate a reduced size image, balancing or improving decoding speed and decoded image quality, by the operations corresponding to the above-described embodiment. Further, this recording medium makes it possible to provide a system that can generate a reduced size image, balancing or improving decoding speed and decoded image quality, with the configurations corresponding to the above-described embodiment. It is apparent that this program and this recording medium can be realized easily based on the later-described embodiment in addition to the above-described embodiment.

A description will be given of a recording medium storing a program or data for realizing the image data generating function of the present invention according to a sixth embodiment of the present invention.

Specifically, the recording medium may be any of a CD-ROM, a magneto-optical disk, a DVD-ROM, a FD, a flash memory, and other various ROMs and RAMs. The recording medium is recorded with a program for causing a computer to execute the functions relating to the above-described embodiment according to the present invention so as to realize the image data generating function of the present invention. By distributing this recording medium, the above-described functions can be realized easily. Then, the image data generating function according to the present invention can be performed by loading the recording medium into an information processing apparatus such as a computer and causing the information processing apparatus to read out the program or by storing the program in a recording medium provided in an information processing apparatus and reading out the program as required.

A specific description will be given of the partial decoding operation of the present invention.

As will be later described, the amount or ratio of codes not to be decoded in the partial decoding operation of the present invention means, for instance, the “number of bit planes not to be decoded” or the “number of bit planes not to be decoded against the total number of coded bit planes.”

Since this embodiment employs an apparatus configuration equal to those described with reference to

In this embodiment, JPEG 2000 selecting the 5×3 wavelet transform is employed. Therefore, a description will first be given of the outline of the JPEG 2000 operation. The application range of the present invention is not limited to JPEG 2000.

Referring again to

Since the color space conversion and inverse conversion unit **21** is previously described, a description will now be given of a reduced size image generating operation according to this embodiment while defining “sub-band” and “decomposition level.”

Since a wavelet transform and an inverse wavelet transform themselves are described with reference to **10**, **11**, and **15**, a description thereof will be omitted.

A description will be given of the encoding of the wavelet coefficients of each sub-band. As is well known, in JPEG 2000, it is possible to code the coefficients from MSB to LSB in bit planes in each sub-band. Strictly, the encoding is performed by subdividing each bit plane, a description of which is omitted.

**2**LL sub-band, and

For purposes of this application, it is supposed that the coefficients of the **2**LL sub-band **5** _{1 }of

In JPEG 2000, each bit plane is individually coded. That is, it is possible to generate codes with the configuration of **2**LL sub-band and end at the **1**HH sub-band. The bit planes are coded in order from the MSB bit plane to the LSB bit plane, and the decoder can recognize the total number of coded bit planes. Therefore, the decoder can easily omit the decoding of the LSB bit plane or the decoding of the two bit planes counted from the LSB bit plane.

Based on the above, a description will be given of the basic flow of a reduced size image generating operation to which the partial decoding of one embodiment of the present invention is applied.

A description will be given, with reference to **13** according to the sixth embodiment of the present invention. Since the image data is compressed by lossless compression, all of the bit planes of all of the sub-bands are coded.

According to the reduced size image generating operation of this embodiment, first, in step S**71** of **72**, (resolution level) i is set to zero (i=0) and r is decreased by one (r=r−1). In steps S**73** through S**75**, i is incremented while i satisfies i≦r. During this process, in step S**74**, the entropy codes of all components are decoded with respect to each resolution level i. Additionally, as stated above, when the aspect ratio of a reduced size image is different from that of the original image, if a desired resolution (1/kh horizontally and 1/kv vertically) with respect to the original image is provided by a user, the value of r is calculated by the equation (7) using the smaller of kh and kv as k. It is clear that the following examples can be modified to the above case easily.

**0** (luminance), component **1** (color difference V), and component **2** (color difference U) of resolution **0** (minimum resolution), components **0**, **1**, and **2** of resolution **1**, components **0**, **1**, and **2** of resolution **2**, . . . . As is well known, in JPEG 2000, the codes of all components can be arranged in the order of resolutions as shown in FIG. **28**. Therefore, it is possible to extract only the wavelet coefficients from the lowest resolution up to a desired resolution with ease (steps S**71** through S**75**).

In this specification, the expression of a 1/3 “resolution” is frequently used, while in JPEG 2000, a quantity called “resolution level” is employed as a quantity similar to “resolution.” The lowest resolution is 0, and the second lowest resolution is 1. **3**LL is resolution level **0**, the resolution level of **3**HL, **3**LH, and **3**HH is resolution level **1**, the resolution level of **2**HL, **2**LH, and **2**HH is resolution level **2**, and the resolution level of **1**HL, **1**LH, and **1**HH is resolution level **3**. Further, the resolution level of **2**LL is resolution level **1**, and the resolution level of **1**LL is resolution level **2**.

The ratio of the resolution of revolution level r to the resolution of an original image depends on the maximum value of the decomposition level. If the maximum value of the decomposition level is d, the resolution of the wavelet coefficients (sub-band) of resolution level r (r>0) with respect to the original image is 1/2^{(d−r+1)}.Therefore, the value of r that should be set in step S**71** in order to generate a 1/k scale image of the original image is given by the above-described equation:

*r=d−*(log_{2} *k−*1) (7)

Accordingly, when the desired resolution (1/k) with respect to the original image is provided by the user, the value of r is calculated by the equation (7) to be set in step S**71**.

Normally, the wavelet coefficients are quantized, and therefore, should be inversely quantized before being subjected to an inverse wavelet transform. Therefore, when the decoding is completed in each resolution level i lower than resolution level r (that is, “NO” in step S**73**), in step S**76**, all of the decoded wavelet coefficients are inversely quantized. Thereafter, in step S**77**, the inversely quantized wavelet coefficients are subjected to an inverse wavelet transform operation. The above-described selected inverse wavelet transform operation may be applied as the inverse wavelet transform operation of step S**77**. Next, in step S**78**, it is determined whether i equals r. If i does not equal r in step S**78**, in step S**79**, thinning-out processing is performed. In the case of a plurality of components, when i equals r (that is, “YES” in step S**78**) or after thinning-out processing (step S**79**), in step S**80**, inverse color conversion is performed and the operation ends.

Partial decoding, which is one feature of one embodiment of the present invention, is the operation of step S**74**. According to this embodiment, the wavelet coefficients of each component are obtained by partially decoding the codes of each component in the following manner, depending on the value of i of the resolution level (or depending on the decomposition level).

Referring to **81**, it is determined whether the component is the luminance Y. If the component is other than the luminance Y, the operation of **32** is started. With respect to the luminance Y (that is, in the case of “YES” in step S**81**), in step S**82**, it is determined whether i<r−2 is satisfied, and in step S**84**, it is determined whether i≦r−1 is satisfied. Based on the results of determination of these steps, the codes of the luminance Y are partially decoded so that the wavelet coefficients of the luminance Y are obtained.

When i≦r−2 (that is, “YES” in step S**82**), in step S**83**, the codes of all of the bit planes are decoded with respect to each sub-band. Since the LL sub-band exists only at resolution level **0**, all of the codes of the LL sub-band are decoded as a result of step S**83**, so that the amount of codes not to be decoded is reduced.

When r−2<i≦r−1 (that is, “NO” in step S**82** and “YES” in step S**84**), in step S**85**, decoding is performed on the bit planes of the sub-bands except for the two bit planes counted from the LSB bit plane (that is, the LSB bit plane and the next low-order bit plane) with respect to the HL and LH sub-bands and the three bit planes counted from the LSB bit plane with respect to the HH sub-band.

When r−1<i≦r (that is, “NO” in step S**82** and “NO” in step S**84**), in step S**86**, decoding is performed on the bit planes of the sub-bands except for the three bit planes counted from the LSB bit plane with respect to the HL and LH sub-bands and the four bit planes counted from the LSB bit plane with respect to the HH sub-band.

With respect to the codes of the color difference Cr, in step S**91** of **30**) is started. With respect to the color difference Cr (that is, in the case of “YES” in step S**91**), in step S**92**, it is determined whether i≦r−3 is satisfied, in step S**94**, it is determined whether i≦r−2 is satisfied, and in step S**96**, it is determined whether i≦r−1 is satisfied. Based on the results of determination of these steps, the codes of the color difference Cr are partially decoded so that the wavelet coefficients of the color difference Cr are obtained. With respect to the codes of the color difference Cr, the wavelet coefficients are obtained by further omitting the decoding of one more bit plane than with respect to the codes of the luminance Y.

When i≦r−3 (that is, “YES” in step S**92**), in step S**93**, the codes of all of the bit planes are decoded with respect to each sub-band.

When r−3<i≦r−2 (that is, “NO” in step S**92** and “YES” in step S**94**), in step S**95**, decoding is performed on the bit planes of the sub-bands except for the two bit planes counted from the LSB bit plane with respect to the HL and LH sub-bands and the three bit planes counted from the LSB bit plane with respect to the HH sub-band.

When r−2<i≦r−1 (that is, “NO” in step S**92**, “NO” in step S**94**, and “YES” in step S**96**), in step S**97**, decoding is performed on the bit planes of the sub-bands except for the three bit planes counted from the LSB bit plane with respect to the HL and LH sub-bands and the four bit planes counted from the LSB bit plane with respect to the HH sub-band.

When r−1<i≦r (that is, “NO” in steps S**92**, S**94**, and S**96**), in step S**98**, decoding is performed on the bit planes of the sub-bands except for the four bit planes counted from the LSB bit plane with respect to the HL and LH sub-bands and the five bit planes counted from the LSB bit plane with respect to the HH sub-band.

With respect to the codes of the color difference Cb, in step S**101** of **31** is started. With respect to the color difference Cb (that is, in the case of “YES” in step S**101**), in step S**102**, it is determined whether i≦r−3 is satisfied, in step S**104**, it is determined whether i≦r−2 is satisfied, and in step S**106**, it is determined whether i≦r−1 is satisfied. Based on the results of determination of these steps, the codes of the color difference Cb are partially decoded so that the wavelet coefficients of the color difference Cb are obtained. With respect to the codes of the color difference Cb, the wavelet coefficients are obtained by further omitting the decoding of one more bit plane than with respect to the codes of the color difference Cr.

When i≦r−3 (that is, “YES” in step S**102**), in step S**103**, the codes of all of the bit planes are decoded with respect to each sub-band.

When r−3<i≦r−2 (that is, “NO” in step S**102** and “YES” in step S**104**), in step S**105**, decoding is performed on the bit planes of the sub-bands except for the three bit planes counted from the LSB bit plane with respect to the HL and LH sub-bands and the four bit planes counted from the LSB bit plane with respect to the HH sub-band.

When r−2<i≦r−1 (that is, “NO” in step S**102**, “NO” in step S**104**, and “YES” in step S**106**), in step S**107**, decoding is performed on the bit planes of the sub-bands except for the four bit planes counted from the LSB bit plane with respect to the HL and LH sub-bands and the five bit planes counted from the LSB bit plane with respect to the HH sub-band.

When r−1<i≦r (that is, “NO” in steps S**102**, S**104**, and S**106**), in step S**108**, decoding is performed on the bit planes of the sub-bands except for the five bit planes counted from the LSB bit plane with respect to the HL and LH sub-bands and the six bit planes counted from the LSB bit plane with respect to the HH sub-band.

Thus, Y<Cr<Cb holds among the components in terms of the amount of codes not to be decoded. Further, the amount of codes not to be decoded is smaller at a higher decomposition level. When each sub-band has the same total number of coded bit planes, the amount of codes not to be decoded is directly equal to the ratio of codes not to be decoded.

In the case of JPEG 2000, an MQ-coder is employed as an entropy coder. In the partial decoding of one embodiment of the present invention, it is important that the code configuration allow partial omission of decoding. Since the details of the entropy decoder have no direct relation to that nature of the partial decoding of one embodiment of the present invention, a description thereof will be omitted.

Referring back to **78** (that is, “YES” in step S**78**), the resolution specified by the user matches the resolution level of wavelet transform. Therefore, after performing an inverse quantization and an inverse wavelet transform, an inverse color conversion is performed without thinning-out processing so that a desired reduced size image is generated. On the other hand, if i=r in step S**78** (that is, “NO” in step S**78**), the resolution level of wavelet transform is higher than the resolution specified by the user. Therefore, in step S**79**, each of Y, Cr, and Cb is converted to a desired size by well-known thinning-out processing after being subjected to an inverse quantization and an inverse wavelet transform. In this embodiment, in order to reduce the amount of processing, thinning-out processing is performed before the inverse color conversion instead of thinning out the R, G, and B values obtained after the inverse color conversion.

Omitting the decoding of the LSB bit plane is equivalent to dividing (quantizing) the wavelet coefficients by two. Omitting the decoding of the n bit planes counted from the LSB bit plane is equivalent to quantizing the wavelet coefficients by 2^{n}. Therefore, in the case of omitting the decoding of bit planes, multiplying the decoded wavelet coefficients by 2^{n }is equivalent to performing an inverse quantization on the decoded wavelet coefficients.

In this embodiment, the equations (8) are employed to calculate Y, Cr, and Cb. In the case of calculating Y, Cr, and Cb using the well known equations provided by CCIR 601, however, the technique of this embodiment is also applicable. Further, the nearest neighbor method, which is widely used, is taken as an example of thinning-out processing. Since the algorithm of thinning-out processing itself does not affect the partial decoding of one embodiment of the present invention, an explanation of the algorithm will be omitted. For the details of the algorithm, see “Introduction to Computer Image Processing” (by Hideyuki Tamura, Soken Publishing Ltd., 1985).

In the above-described case, the conditional expression of step S**73** is i≦r. Therefore, the resolution level of the extracted wavelet coefficients is equal to or higher by one than the resolution specified by the user. If the conditional expression of step S**73** is changed to i<r, however, the resolution level lower than the resolution level specified by the user by one can be extracted, so that each of Y, Cr, and Cb is converted to a desired size by well known interpolation processing after being subjected to an inverse wavelet transform.

According to the reduced size image generating operation of **111**, the resolution level of a reduced size image to be generated is set to (resolution level) r based on a resolution (1/k) specified by the user. Then, in step S**112**, (resolution level) i is set to zero (i=**0**) and r is decreased by one (r=r−1). In steps S**113** through S**115**, i is incremented while i satisfies i≦r. During this process, in step S**114**, the entropy codes of all components are decoded with respect to each resolution level i. As in the above-described case, the operations described with reference to **114**.

Normally, the wavelet coefficients are quantized, and therefore, should be inversely quantized before being subjected to an inverse wavelet transform. Therefore, when the decoding is completed in each resolution level i lower than resolution level r (that is, “NO” in step S**113**), in step S**116**, all of the decoded wavelet coefficients are inversely quantized. Thereafter, in step S**117**, the inversely quantized wavelet coefficients are subjected to an inverse wavelet transform operation. The above-described selected inverse wavelet transform operation may be applied as the inverse wavelet transform operation of step S**117**. Next, in step S**118**, interpolation is performed. The nearest neighbor method, which is widely used, is taken as an example of interpolation. For the expatiation of the algorithm of interpolation, the above-described “Introduction to Computer Image Processing” may be referred to. In the case of a plurality of components, in step S**119**, an inverse color conversion is performed and the operation ends.

Thus, according to one embodiment of the present invention, the wavelet coefficients are subjected to an inverse wavelet transform, and in the case of generating a reduced size image of an original image, the wavelet coefficients in the same sub-band are selectively subjected to an inverse wavelet transform. Therefore, the reduced size image can be generated with a reduced amount of calculation.

Further, according to one embodiment of the present invention, the wavelet coefficients in the same sub-band are partially decoded in performing an inverse wavelet transform on the wavelet coefficients and generating a reduced size image of an original image. Therefore, the reduced size image can be generated with decoding speed and decoded image quality being balanced with each other.

The present invention is not limited to the specifically disclosed embodiments, but variations and modifications may be made without departing from the scope of the present invention.

The present application is based on Japanese priority application No. 2002-072696 filed on Mar. 15, 2002, the entire contents of which are hereby incorporated by reference.

Whereas many alterations and modifications of the present invention will no doubt become apparent to a person of ordinary skill in the art after having read the foregoing description, it is to be understood that any particular embodiment shown and described by way of illustration is in no way intended to be considered limiting. Therefore, references to details of various embodiments are not intended to limit the scope of the claims which in themselves recite only those features regarded as essential to the invention.

Patent Citations

Cited Patent | Filing date | Publication date | Applicant | Title |
---|---|---|---|---|

US5563960 | Dec 17, 1993 | Oct 8, 1996 | David Sarnoff Research Center, Inc. | Apparatus and method for emphasizing a selected region in the compressed representation of an image |

US6041143 | Apr 14, 1998 | Mar 21, 2000 | Teralogic Incorporated | Multiresolution compressed image management system and method |

US6101284 * | Oct 28, 1996 | Aug 8, 2000 | Ricoh Company, Ltd. | Methods and systems for optimizing image data compression involving wavelet transform |

US6137595 | Apr 16, 1998 | Oct 24, 2000 | Ricoh Company, Ltd. | Color conversion processing method for color matching and a processor readable medium storing a program for implementing the color conversion processing method |

US6215916 | Feb 4, 1998 | Apr 10, 2001 | Intel Corporation | Efficient algorithm and architecture for image scaling using discrete wavelet transforms |

US6226011 | Mar 31, 1998 | May 1, 2001 | Ricoh Company, Ltd. | Color conversion method for a color matching, a method for expanding a function of a program module and a color conversion method using the method for expanding a function of a program module |

US20020039440 | Jul 26, 2001 | Apr 4, 2002 | Ricoh Company, Ltd. | System, method and computer accessible storage medium for image processing |

US20020159644 | Feb 26, 2002 | Oct 31, 2002 | Hiroyuki Sakuyama | Image processing scheme |

US20020196970 | May 29, 2002 | Dec 26, 2002 | Yutaka Sano | Image compression apparatus, image decompression apparatus, image compression method, image decompression method, computer program product, and recording medium |

US20030002742 | Jun 10, 2002 | Jan 2, 2003 | Yutaka Sano | Image compression method and apparatus, image expansion method and apparatus, and storage medium |

US20030179941 * | Feb 25, 2003 | Sep 25, 2003 | Decegama Angel | Methods for real-time software video/audio compression, transmission, decompression and display |

EP0859335A2 | Jan 7, 1998 | Aug 19, 1998 | Canon Kabushiki Kaisha | Thumbnail image manipulation using aspect ratio zooming, compression and scaling |

EP1349393A1 | Mar 11, 2003 | Oct 1, 2003 | Ricoh Company | Image compression device, image decompression device, image compression/decompression device, program for executing on a computer to perform functions of such devices, and recording medium storing such a program |

FR2805640A1 | Title not available | |||

JP2000040142A | Title not available | |||

JP2000125293A | Title not available | |||

JP2000125294A | Title not available | |||

JP2001285642A | Title not available | |||

JP2002152517A | Title not available | |||

JP2002152744A | Title not available | |||

JP2002344732A | Title not available | |||

JPH0512800A | Title not available | |||

JPH0564001A | Title not available | |||

JPH0622152A | Title not available | |||

JPH0644692A | Title not available | |||

JPH06350989A | Title not available | |||

WO1999049412A1 | Sep 15, 1998 | Sep 30, 1999 | Univ Maryland | Method and apparatus for compressing and decompressing images |

Non-Patent Citations

Reference | ||
---|---|---|

1 | E. Nguyen et al., "A ROI Approach for Hybrid Image Sequence Coding", Proceedings of the International Conference on Image Processing (ICIP), Austin, Nov. 13-16, 1994, Los Alamitos, vol. 3, Conf. 1, pp. 245-249. | |

2 | Hideyuki Tamura, "Introduction to Computer Image Processing", Soken Publishing, Ltd., 1985, 7 pages. | |

3 | Joel Askelof et al., "Region of Interest Coding in JPEG 2000", Signal Processing: Image Communication, vol. 17, No. 1, Jan. 2002, pp. 105-111. | |

4 | JPEG 2000 Part I Committee Draft Version 1.0, Coding of Still Pictures, ISO/IEC JTC1/SC29 WG1, Mar. 16, 2000, XP-001004858, pp. 1-190. | |

5 | Panchanathan, S. et al., "Image Scalability Using Wavelet Vector Quantization" Journal of Electronic Imaging, SPIE/IS & T, US, vol. 5, No. 2, Apr. 1996, pp. 167-174, XP000596287. | |

6 | Panchanathan, S. et al., "Wavelet Based Scalable Image Compression", Proceedings of the SPIE, SPIE Bellingham, VA, US, vol. 2419, Feb. 7, 1995, pp. 505-514, XP008006331. | |

7 | Zhong Min Sheng, et al., "JPEG2000-Based Scalable Reconstruction of Image Local Regions", Proceedings of Int'l. Symposium on Intelligent Multimedia, Video and Speech Processing, 2001, pp. 174-177, XP002211321. |

Referenced by

Citing Patent | Filing date | Publication date | Applicant | Title |
---|---|---|---|---|

US7382922 * | Jun 24, 2004 | Jun 3, 2008 | Canon Kabushiki Kaisha | Creation of remarkable regions in a digital image |

US7885486 * | Jun 1, 2007 | Feb 8, 2011 | Fuji Xerox Co., Ltd. | Image processing system, method for processing image and computer readable medium |

US7912324 | Apr 28, 2006 | Mar 22, 2011 | Ricoh Company, Ltd. | Orderly structured document code transferring method using character and non-character mask blocks |

US8649616 | Sep 8, 2010 | Feb 11, 2014 | Ricoh Company, Limited | Image processing apparatus and image processing method |

US20050013493 * | Jun 24, 2004 | Jan 20, 2005 | Canon Kabushiki Kaisha | Creation of remarkable regions in a digital image |

US20060062554 * | Sep 24, 2003 | Mar 23, 2006 | Jupiter Jay A | Picture viewer |

US20060130103 * | Oct 12, 2005 | Jun 15, 2006 | Matsushita Electric Industrial Co., Ltd. | Video playback device |

US20080025643 * | Jun 1, 2007 | Jan 31, 2008 | Fuji Xerox Co., Ltd. | Image processing system, method for processing image and computer readable medium |

Classifications

U.S. Classification | 382/240, 358/426.14, 348/398.1 |

International Classification | G06K9/46, G06T3/40, H04N11/02, H04N1/32, H04N7/12, H04N1/41, G06K9/36 |

Cooperative Classification | G06T3/4084 |

European Classification | G06T3/40T |

Legal Events

Date | Code | Event | Description |
---|---|---|---|

Jun 2, 2003 | AS | Assignment | Owner name: RICOH CO., LTD., JAPAN Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:SAKUYAMA, HIROYUKI;REEL/FRAME:014131/0503 Effective date: 20030509 |

Nov 30, 2010 | FPAY | Fee payment | Year of fee payment: 4 |

Nov 27, 2014 | FPAY | Fee payment | Year of fee payment: 8 |

Rotate